Winning Solution in NTIRE19 Challenges on Video Restoration and Enhancement (CVPR19 Workshops) - Video Restoration with Enhanced Deformable Convolutional Networks. EDVR has been merged into BasicSR and this repo is a mirror of BasicSR.
Hi @xinntao
Thanks for your great work and fancy idea in CVPR19W paper. I have a question when I try to reproduce your moderate model work for less training time.
Test results:
I think the data of 25.83/0.8077 is the test results of Vimeo__M model in Table 4 right.
And how many iters do your train ? Is 600k ?
train_EDVR_woTSA and train_EDVR_M model of Vimeo dataset
Hi @xinntao Thanks for your great work and fancy idea in CVPR19W paper. I have a question when I try to reproduce your moderate model work for less training time.
Test results: I think the data of 25.83/0.8077 is the test results of Vimeo__M model in Table 4 right.
And how many iters do your train ? Is 600k ?
train_EDVR_woTSA and train_EDVR_M model of Vimeo dataset
There are only Vimeo_L , train_REDS_woTSA_M and train_REDS_M pretrained models on the below two links. https://github.com/xinntao/EDVR/wiki/Model-Zoo https://github.com/xinntao/EDVR/wiki/Testing-and-Training
Could you provide models of Vimeo_woTSA_M and Vimeo_M ?